Apriel-H1: Towards Efficient Enterprise Reasoning Models
1–2 of 2 posts
Re: Apriel-H1: Towards Efficient Enterprise Reasoning Models
#2Apriel-H1-15b-Thinker-SFT uses incremental distillation from Apriel-Nemotron-15B-Thinker, selectively replacing less critical attention layers with linear Mamba blocks to reduce computational complexity while preserving reasoning quality.